Transcription fails
Symptoms
Section titled “Symptoms”A session enters an error state during transcription, or no usable transcript appears after conversion completes.
Likely causes
Section titled “Likely causes”The selected Whisper model may be unavailable, the converted audio may be unreadable, the bundled whisper.cpp binary may be unavailable, or the source has little usable speech.
Fast checks
Section titled “Fast checks”Read the session error, confirm the selected Whisper model is installed, and distinguish a preceding conversion error from a transcription error.
Resolution
Section titled “Resolution”Retry only after the missing model or conversion problem is resolved. Keep the original source and test a known-good, non-sensitive recording if you need to separate source-media behavior from runtime behavior.
Data-loss warning
Section titled “Data-loss warning”Do not delete the failed session or source recording before preserving its exact error. Resetting data prevents comparison with the original failure.
Information to collect for support
Section titled “Information to collect for support”Collect the session status, full error, selected model and language, media type, whether conversion completed, and the application and macOS versions.